Prediction of jet noise from circular beveled nozzles

In this paper, simulations are performed of flow-induced noise from circular beveled jet nozzles. The Detached-Eddy Simulation (DES) approach is used to simulate both the jet nozzle internal and external flows as well as the jet plume. This methodology allows the turbulence model to transition from an unsteady Reynolds Averaged Navier-Stokes (URANS) method for attached boundary layers to a Large-Eddy Simulation (LES) in separated regions. A cylindrical coordinate system is used, with the centerline of the jet nozzle coinciding with the coordinate system's polar axis. The one equation Spalart-Allmaras turbulence model is used to describe the evolution of the turbulent eddy viscosity. An explicit 4th order Runge-Kutta time marching scheme is used. To avoid the use of a full turbulent grid in the nozzle, simple wall functions are used. The far-field sound is evaluated using the Ffowcs Williams-Hawkings permeable surface acoustic analogy. The computational domain is divided into at least two blocks, with the nozzle being along part of the boundary of the inner and outer blocks. A rigid cylindrical boundary, with finite thickness, is used to represent the jet exhaust nozzle. A nozzle with uniform inlet flow of M = 0.9, for both unheated and heated jets is considered. The grid distribution is uniform in the axial and azimuthal directions. The grid is clustered in the radial direction near the lip line of the jet. It is observed that the thrust axis deviates from the geometric centerline at an angle of 8.5 degrees. This is in reasonable agreement with experimental flow measurements. Flow characteristics such as the variation of the mean velocity and turbulent intensity in the radial, axial and azimuthal directions are described. The Ffowcs Williams-Hawkings calculations are performed using the code PSU-WOPWOP developed at Penn State. For these calculations, the surface chosen for the beveled nozzle case is a conical surface at r = 6D and length x = 25D, with a cone angle equal to the deflection angle of the thrust axis. The predictions capture the peak noise levels above and below the bevel as well as the crossover in the spectra at lower polar angles to the jet exit and the shielding effect at higher angles. Noise characteristics for the beveled nozzle are compared with experimental measurements. In addition, the effect of, and need for, artificial excitation to stimulate the development of a turbulent jet mixing layer are studied. Finally, the choice of acoustic data surface for the Ffowcs Williams-Hawkings permeable surface method, as well as whether it should be closed or open at the downstream end are considered.